Microsoft Word - Pallotti & Borghetti pp.1-20 4Ed final.doc EuroAmerican  Journal  of  Applied  Linguistics  and  Languages   Special  Issue,  Volume  6,  Issue  1,  August  2019,  1-­‐20   ISSN  2376-­‐905X   DOI:  http://dx.doi.org/10.21283/2376905X.10.162     www.e-­‐journall.org       ©  Pallotti  2019.  This  work  is  licensed  under  a  Creative  Commons  Attribution  4.0  International  License.       The  effects  of  an  experimental  approach  to  writing  instruction  on   monolingual  and  multilingual  pupils  in  Italian  primary  schools         GABRIELE  PALLOTTIa  &  CLAUDIA  BORGHETTIb   aUniversità  di  Modena  and  Reggio  Emilia   bUniversità  di  Bologna       Received  6  December  2018;  received  in  revised  form  24  May  2019;  accepted  8  July  2019 ABSTRACT EN This article reports on the results of an action-research project in Italian primary schools. Its main principles are formative evaluation, a process orientation to writing, cooperative learning, and an inclusive approach that does not separate pupils according to their linguistic or writing skills. Pupils aged eight to eleven years old participated in the study in seven experimental (N = 106) and seven control classes (N = 118). Their texts were examined both holistically, through scales of communicative effectiveness, and by means of analytical measurements. The comparison of monolingual and multilingual pupils in both experimental and control classes shows a number of similarities in terms of textual competence, as well as a few differences, especially regarding the more linguistic aspects of writing, such as verb tense cohesion. A remarkable result is that students’ text quality seems to depend more on their participation to the program than on their monolingual or multilingual status. Key words: MULTILINGUAL PUPILS, L1/L2 WRITING, WRITING DEVELOPMENT, WRITING INSTRUCTION, PRIMARY SCHOOL ES Este artículo informa los resultados de un proyecto de investigación-acción en escuelas primarias italianas. Sus principios más importantes son la evaluación formativa, una orientación al proceso de escritura, aprendizaje cooperativo y un enfoque inclusivo que no separa a los estudiantes según sus habilidades lingüísticas o de escritura. Los alumnos de entre ocho y once años participaron en el estudio en siete clases experimentales (N = 106) y siete clases control (N = 118). Sus textos se examinaron de manera holística por medio de escalas de efectividad comunicativa y mediciones analíticas. La comparación de alumnos monolingües y multilingües en las clases experimentales y de control muestra un número de similitudes en términos de competencia textual al igual que algunas diferencias, especialmente con respecto a los aspectos más lingüísticos de la escritura, como la cohesión del tiempo verbal. Un resultado notable muestra que participar en el programa tuvo un mayor impacto en la calidad del texto de los estudiantes que sus condiciones iniciales mono- o multilingüe. Palabras clave: ALUMNOS MULTILINGÜES, ESCRITURA L1/L2, DESARROLLO DE ESCRITURA, INSTRUCCIÓN DE ESCRITURA, ESCUELA PRIMARIA IT Questo articolo dà conto dei risultati di un progetto di ricerca-azione in scuole primarie italiane. I principi centrali sono la valutazione della formazione, l’orientamento al processo della scrittura, l'apprendimento cooperativo e l’approccio inclusivo che non separa gli/le apprendenti in base alle loro capacità linguistiche o di scrittura. A questo studio hanno partecipato apprendenti di età compresa tra gli 8 e gli 11 anni, in sette classi sperimentali (N = 106) e in sette classi di controllo (N = 118). I loro testi sono stati esaminati sia olisticamente attraverso scale di efficacia comunicativa, sia mediante misurazioni analitiche. Il confronto tra alunni monolingui e multilingui, sia nelle classi sperimentali sia in quelle di controllo, mostra una serie di similarità in termini di competenza testuale e anche alcune differenze, in particolare per quanto riguarda gli aspetti più linguistici della scrittura, come la coesione verbale. Un risultato importante mostra che la partecipazione al programma ha avuto un impatto maggiore sulla qualità del testo degli/delle apprendenti rispetto alle loro condizioni iniziali di monolingui o multilingui. Parole chiave: APPRENDENTI MULTILINGUI, SCRITTURA L1/L2, SVILUPPO DELLA SCRITTURA, INSEGNAMENTO DELLA SCRITTURA, SCUOLA PRIMARIA   *  Gabriele  Pallotti,  Università  di  Modena  e  Reggio  Emilia                gabriele.pallotti@unimore.it   PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 2   1.  Introduction   In   the   field   of   teacher   education,   literacy   and   second   language   acquisition   (SLA)   are   often   kept   separate.   In   teaching  practice   too,  actions  aimed  at   improving  writing  skills  and  actions  stimulating  second   (or  additional)  language  acquisition  are  at  times  seen  as  complementary,  but  substantially  distinct,  based  on   different   theoretical  and  methodological  assumptions  and  approaches.  This  article  reports   the  results  of  an   action-­‐research  teacher-­‐training  project  intended  to  help  bridge  these  gaps,  by  adopting  an  approach  to  the   teaching   of   writing   that   involves   all   students,   regardless   of   whether   they   are   using   their   first,   second,   or   additional   language.   More   specifically,   the   context   of   the   study   was   a   teacher-­‐training   initiative  Osservare   l’interlingua  whose   aim  was   to   promote   effective   language   education,  with   special   emphasis   on  writing,   at   elementary  and  middle  school  levels.     Writing  is  a  complex  activity,  involving  several  sub-­‐skills  (e.g.,  sorting  through  one’s  memory  of  the   events/facts,   organizing   them,   selecting   appropriate   language   resources),   which   may   be   more   or   less   developed   in   different   people.   This   complexity   is   greater   in  multilingual   speakers  with   variable   control   of   different   linguistic   codes   and   their   sub-­‐domains,  which   in   turn  may   be  more   or   less   implicated   in  written   communication.   Teachers   need   to   be   able   to   take   apart   and   carefully   analyze   all   these   factors,   in   order   to   allow   for   them   in   their  daily  practices.   In  most  cases,   these  daily  practices  should  be  directed  at   the  whole   class,   rather   than   specific   sub-­‐groups  made   up   of  monolingual   or  multilingual   students,   or   first   or   second   language  learners,  with  higher  or  lower  writing  skills.  This  article  reports  on  a  program  that  attempts  to  meet   these  objectives,  and  whose  effectiveness  will  be  assessed  using  a  quasi-­‐experimental  research  design.   Before  proceeding  with  the  exposition,  some  terminological  clarification  is  in  order.  The  term  second   language  (or  L2)  has  been  repeatedly  called  into  question  in  recent  years.  Firstly,  in  many  cases  the  language   at   stake   is   not   the   second   a   person   is   acquiring,   but   often   a   third,   fourth,   or   further   language.   This   is  why   additional   language   acquisition   has   been   proposed   as   a   more   correct   replacement   for   second   language   acquisition,   and   LX   as   a   more   general   term   than   L2   (Dewaele,   2017).   As   a   consequence,   many   so-­‐called   bilinguals  should  actually  be  called  multilingual  speakers.  This  term  multilingual  also  overcomes  the  limits  of   expressions   such   as  non-­‐native   speaker   or  L2   learner,   which   imply   some   sort   of   deficit  with   respect   to   an   idealized  native  speaker  who  is  treated  as  the  necessary  benchmark  of  language  performance  (Cook,  1999).   Secondly,  the  simple  L1/L2  dichotomy  does  not  apply  to  other  groups  of  learners,  such  as  so-­‐called  heritage   speakers,  i.e.,  children  of  immigrants  who  are  either  monolingual  or  dominant  in  the  language  of  the  family  in   early   childhood,   but   progressively   develop   a   balanced   bilingualism   or   eventually   become   dominant   in   the   majority  language  (Montrul,  2016). This   article   adopts   solely   the   distinction   between   monolingual   and   multilingual   students   (except   when  citing  research  that  explicitly  refers  to  “L2  acquisition”  and  “bilingualism”).  This  categorisation  seems   relatively   descriptive   and   neutral:   it   does   not   require   one   to   draw   the   problematic   line   between   first   and   second  language  acquisition,  or  between  native  and  non-­‐native  speakers,  and  it  does  not  assume  any  deficit   view  of  language  learning  and  use,  but  rather  describes  a  person’s  linguistic  repertoire  as  being  composed  of   one  or  more  codes  employed  in  everyday  communication.1       2.  Writing  development  and  instruction:  an  overview   2.1.  Writing  in  L1  and  LX   The  practices  of  L1  and  LX  writing  have  frequently  been  compared,  both  in  terms  of  processes  (i.e.,   the   cognitive   activities   employed   when   writing)   and   in   terms   of   products   (the   texts).   Analysing   the   interaction  of  four  cognitive  activities  implied  in  writing  (reading  the  assignment,  planning,  generating  ideas,   and  formulating),  van  Weijen  (2009)  concludes  that  the  distribution  of  each  activity  is  similar  in  both  L1  and   L2  writing.  These  results  confirm  many  previous  studies  (Cumming,  Rebuffot  &  Ledwell,  1989;  Uzawa,  1996;   for   a   review:   Cumming,   2001).   However,   a   consistent   number   of   differences   between   L1   and   LX   cognitive   activities   have   also   been  highlighted   in   text   planning   and   revision,   and   in  writing   fluency   (Jones  &  Tetroe,   1987;  Hall,  1990;  Stevenson,  Schoonen,  &  de  Glopper,  2006;  van  Waes  &  Leijten,  2015).  This  overall  empirical   evidence   (e.g.,   Leki,   Cumming,  &   Silva,   2006;   Silva,   1993)   tends   to   converge   into   a   general   agreement   that   writing  processes  in  L1  and  LX  are  similar,  even  if  “L2  composing  is  more  constrained,  more  difficult,  and  less   1  The  multilingual  category  does  not  include  those  children  who  only  know  a  smattering  of  a  foreign  language  taught  at   school  in  addition  to  their  first  language.   AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 3   effective”  (Silva,  1993,  p.  668),  mainly  because  LX  writers  need  to  concentrate  on  language  forms  as  well  as   on  content.   Contrasting   conclusions   have   also   been   drawn   from   research   which   compares   the   linguistic   properties   of   texts  produced  by  L1   and  LX  writers.   For   example,   Silva   (1993)   claims   that   L2  writers’   texts   contain  fewer  words,  while  Cameron  and  Besser  (2004)  point  out  that  text  length  does  not  differ  significantly,   even   if   L2  writers   tend   to  make   lesser  use  of   adverbials,  which   in   turn   reduces   the  amount  of   information   they  provide  about  the  story  (e.g.,  in  terms  of  time,  place  and  manner  of  the  events).   Some   research   has   also   highlighted   that   LX   writers   employ   a   restricted   repertoire   of   features   (especially  conjunctions  and  demonstrative  pronouns)  to  give  cohesion  to  their  texts  (Hinkel,  2001;  see  also:   Oi,   1984   and  Reid,   1988,   both   cited   in   Silva,   1993).  However,   less   is   known   about   how  L1   and   LX  writers   manage  reference  to  entities,  that  is,  which  linguistic  and  discourse  features  they  use  to  make  the  characters   of  their  stories  identifiable  by  the  reader.  While  it  has  been  claimed  that  L1  children  learn  to  systematically   use  appropriate  referent  introductions  from  seven  years  onwards  (Hickmann,  1995),  little  or  no  research  has   been  conducted,  to  the  best  of  our  knowledge,  on  the  developmental  patterns  of  young  LX  writers,  and  one  of   the   aims   of   the   present   article   is   to   start   filling   this   gap.   Hendriks’s   (1998)   study,   comparing   the   written   productions  of  adult  LX   learners  with  those  of  L1  children,  may  in  fact  be  more  informative   in  terms  of  the   impact  of  age—rather  than  “native”  status—on  cohesion  skills  and/or  preferences.   2.2.  Research-­‐based  writing  instruction   A   considerable   body   of   research   has   accumulated   on   the   effectiveness   of   instruction   on   writing   quality  (e.g.,  Graham,  Harris,  &  Santangelo,  2015;  Rietdijk,  Janssen,  van  Weijen,  van  den  Bergh,  &  Rijlaarsdam,   2017),   and   meta-­‐analyses   are   also   available   (e.g.,   Graham,   Harris,   &   Chambers,   2016),   some   of   which   specifically   focus   on   primary   school   children   (Graham,   McKeown,   Kiuhara,   &   Harris,   2012;   Koster,   Tribushinina,   de   Jong,   &   van   den   Bergh,   2015).   These   last   authors   identified   a   number   of   instructional   practices  that  have  systematically  produced  positive  results  with  younger  learners.  These  include:     − Teaching   writing   strategies   for   planning,   drafting,   or   revising   one’s   text,   e.g.,   teaching   pupils   to   organize   ideas   in  writing  plans  and  to  use/upgrade   the  plan  when  writing  (Tracy,  Reid,  &  Graham,   2009).   − Assigning   specific   goals   before   and   during   the   writing   process,   like   asking   students   to   add   information  about  a  specific  fact/character,  to  divide  the  text  into  paragraphs,  or,  as  Ferretti,  Lewis,   and   Andrews-­‐Weckerly   (2009)   have   shown,   to   follow   some   given   prompts  when   organising   their   argumentative  discourse.   − Promoting   group  work   and   peer   collaboration,   as   pupils   plan,   draft,   and   revise   their   text;   see,   for   example,  Brakel  Olson  (1990)  about  peer-­‐to-­‐peer  support  in  revision.   − Teaching   text   structure  explicitly,  e.g.,  how  narrative   texts  work   in   terms  of   story  constituents  and   event  sequences  (Fitzgerald  &  Teasley,  1986). − Providing  teacher  and  peer-­‐to-­‐peer  feedback  during  and  after  writing  (e.g.,  Schunk  &  Swartz,  1993).     Not   surprisingly,   these   research-­‐based   best   practices   in   L1   writing   instruction   are   also   often   mentioned   in   relation   to  LX  writing.   For   example,  Olson,   Scarcella,   and  Matuchniak   (2013)   suggest   that,   in   order   to   limit   the   effect   of   LX   writers’   cognitive   constraints,   educators   should   focus   on   the   teaching   of   strategies  which  help  students  set  intermediate  goals  and  thus  manage  the  complexity  of  the  writing  process.   Moreover,  LX  collaborative  writing  and  peer-­‐to-­‐peer  feedback  are  often  advocated  (Storch,  2013),  as  well  as   genre  pedagogy  (Hyland,  2007).   However,  research  to  date  on  the  (differential)  effects  of  pedagogical   interventions  on  monolingual   and   multilingual   pupils’   writing   remains   scant.   This   is   one   of   the   motivations   behind   this   study,   whose   pedagogical  approach  largely  draws  on  the  works  briefly  reviewed  in  this  section.   3.  The  study   3.1.  Research  questions   The   following   research   questions   were   formulated   in   order   to   investigate   the   outcomes   of   an   inclusive  writing  program  on  monolingual  and  multilingual  learners:   PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 4   − RQ1.  In  a  primary  school  context,  what  are  the  effects  on  the  whole  class  of  a  long-­‐term  pedagogical   intervention   based   on   the   process   approach   to   writing,   formative   assessment,   and   cooperative   learning? − RQ2.  If  there  are  any  effects,  do  they  differ  for  monolingual  and  multilingual  pupils? − RQ3.  What   are   the   differences   between  monolingual   and  multilingual   pupils   in   control   classes   as   regards  writing  skills?   3.2.  The  project     The   project  Osservare   l'interlingua,   promoted   by   the   Reggio   Emilia  municipality   together  with   the   University   of  Modena   and   Reggio   Emilia   (interlingua.comune.re.it;   Pallotti,   2010,   2017a,   2017b;   Pallotti   &   Rosi,   2017)   started   in   Reggio   Emilia   (Italy)   in   2007.   It   involves   12   elementary   and   3   middle   schools,   28   teachers  and  35  classes,  for  a  total  of  about  650  pupils  aged  between  6  and  14  (Grades  1  to  8).  While,  at  the   beginning,   it  was  meant  to  address  issues  of  LX  Italian  learning,  today  it   is  directed  toward  all  the  pupils   in   class,  with  an   inclusive  approach   involving  monolingual  and  multilingual   students  with  varying  proficiency   levels.   The  project  is  both  a  teacher  training  and  action-­‐research  initiative,  with  teachers  actively  involved   in   several   ways.   Its   main   tenet   is   that   effective   teaching   practices   should   be   grounded   in   the   careful   observation  of  students’  needs,  competences,  and  learning  strategies.  Thus,  teachers  are  first  of  all  trained  to   conduct   systematic   analyses   of   their   pupils’   oral   and   written   productions,   in   line   with   the   interlanguage   approach  (Pallotti,  2017b;  Selinker,  1972).  This  is  a  kind  of  formative  assessment  that  aims  to  overcome  the   traditional  stance  of  “hunting  for  errors”  and  to  allow  teachers  to  developing  the  activities  that  best  suit  the   needs  of  their  class.  It  also  promotes  the  learners’  autonomy,  since  they  are  not  seen  as  passive  recipients  of   knowledge   that   has   been   elaborated   elsewhere,   but   rather   as   active   creators   of   their   own   learning.     This   orientation,  and  the  ability  to  analyze  children’s  texts  accordingly,  takes  some  time  to  develop.  It  begins  with   pre-­‐service  training  for  some  of  the  younger  teachers,  while  others  often  acknowledge  that  it  took  them  some   years  to  get  used  to  looking  at  pupils’  productions  in  positive  terms  through  recognizing  their  strategies  and   developments,  rather  than  just  spotting  errors.  This  is  why  text-­‐analysis  sessions  are  periodically  held  (about   once   a   year)   as   part   of   the   teachers’   permanent   training;   in   these   sessions,   more   and   less   experienced   participants  assist  each  other  in  building  their  formative  analytic  skills.   Teachers  also  actively  take  part  in  planning  pedagogical  actions.  At  the  beginning  of  each  school  year,   the   texts   produced   by   pupils   are   analysed   by   the   group   of   teachers   and   researchers.   Considering   the   properties  of  the  texts  and  thus  the  needs  of  each  class,  the  group  of  teachers  defines  a  common  theme  (e.g.,   narrative,  expository,  descriptive   texts)   to  be  pursued  by  all  participants   in   their  classes  and  they  agree  on   common   core   activities.   Then,   each   teacher   introduces   the   due   adaptations   according   to   age   and   overall   proficiency   level   of   their   pupils,   receiving   guidance   from   the   researchers.   The   group   of   teachers   and   university   researchers  meets   approximately   every   two  months   throughout   the   entire   school   year,   so   that   planned  activities  can  be  monitored,  discussed,  and  possibly  revised. The  project  as  a  whole  promotes  a  process  approach  to  teaching  writing  (e.g.,  White  &  Arndt,  1991),   according   to  which   teaching   activities   guide   the   pupils   to   experience   and   progressively   acquire   the   stages   implied   in   text   production:   generating   and   organizing   ideas,  writing   the   initial   draft,   revising,   and   editing.   Most  of  these  activities  are  carried  out  in  a  cooperative  manner,  in  small  and  large  groups  (following  Ede  &   Lunsford,   1990;   Storch,   2013),   and,   in   line   with   socio-­‐cultural   approaches,   they   encourage   the   semiotic   mediation  of  thought  by  making  the  writing  process  more  concrete  through  the  use  of  artefacts  such  as  boxes,   envelopes,   paper   strips,   and   posters   (following   Englert,   Mariage,   &   Dunsmore,   2006).   Feedback,   too,   is   primarily   provided   by   the   students,   as   the   program   particularly   encourages   peer-­‐to-­‐peer   response,   both   within   the  members   of   the   group   and   among   groups   (following   Andrade  &   Evans,   2013).   In   line  with   the   interlanguage   approach,   this   feedback   is   not   summative   (no   grades   are   assigned,   by   either   teachers   or   students),   but   formative,   as   it   focuses   on   pupils’   strengths   and  weaknesses,   analyses   their   strategies,   and   suggests  ways  of  developing  their  skills.     Teachers  are  actively   involved   in  data  collection,   too,  which   is  usually  carried  out  at   the  beginning   and   end   of   every   school   year,   in   relatively   controlled   and   standardized   conditions   (that   are   nonetheless   absolutely   natural   for   a   classroom   setting—no   laboratory   experiments,   in   other  words),   so   as   to   allow   for   comparisons   over   time   and   across   pupils   and   classes.   This   data   is   kept   by   schools   as   a  portfolio   of   pupils’   linguistic  development  over   the  years,  qualitatively  analyzed  by   the   teachers,  and  also  used  by  researchers   AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 5   (and  some  pre-­‐service-­‐training  students  as  part  of  their  master’s  thesis)  for  subsequent  quantitative  analyses,   such  as  the  kind  reported  in  this  article.   3.3.  The  pedagogical  intervention  (experimental  classes)   In   the   course   of   the   2013-­‐2014   school   year,   when   data   collection   for   this   study   took   place,   the   pedagogical   intervention   carried   out   in  most   classes  was   aimed   at   developing   the   pupils’   writing   skills   as   applied  to  narrative  texts.   Although  the  seven  classes  had  slightly  different  needs  and  were  thus  engaged  with  partially  diverse   tasks,  all  of  them  were  involved  in  the  following  core  activities:2     1) Students   wrote   individual   texts   after   watching,   twice,   a   video   clip   of   a   few   minutes   taken   from   Modern  Times   by  Charlie  Chaplin.  This   served   to   collect  baseline  data   to  assess   students’  progress   over  the  school  year  and  to  fine-­‐tune  teaching  activities  based  on  their  competences  and  needs. 2) In  groups,  the  students  viewed  the  initial  video  clip  again,  identified  the  main  sequences  of  the  story   (primary  information  units),  and  labelled  them  with  short  descriptions  written  on  paper  envelopes   or  cardboard  boxes.   3) The   groups   watched   the   video   once   more   and   gathered   additional   information   about   characters,   actions,   etc.   (secondary   information   units),   which   they   noted   on   paper   strips   to   be   placed   in   the   envelopes/boxes.   4) In  their  groups,  pupils  glued  the  strips  on  big  sheets  of  paper,  deciding  on  the  order  to  best  convey   the   story   of   the   video   clip.   This   procedure   led   to   the   text   outline.   Since   a   main   objective   of   the   program  was  to  promote  a  better  organization  of  texts,  special  emphasis  was  placed  on  the  need  to   cluster  similar  strips  of  information  together,  in  order  to  facilitate  the  following  segmentation  of  the   text  into  paragraphs.     5) The  groups  produced  the  first  versions  of  their  texts,  based  on  the  video  clip  they  had  worked  on  in   the  first  four  steps.   6) The  texts  were  then  circulated  among  groups,  who  had  been  previously  instructed  on  how  to  make  a   peer-­‐to-­‐peer  revision  based  on  explicit  and  shared  criteria  (e.g.,  clarity,  lexical  choices,  spelling).  It  is   worth  noting  that  a  number  of  aspects  of  writing,  such  as  verb  tense  cohesion  and  punctuation,  were   addressed   within   this   phase   merely   by   asking   pupils   to   take   them   into   consideration   in   the   reviewing  process.  These  topics  did  not  receive  significant  attention  within  the  other  activities.   7) Each   group   reviewed   their   own   text,   taking   into   account   the   comments   of   the   other   groups.   They   then  rewrote  it,  after  having  to  decide  which  suggestions  for  improvement  they  wanted  to  accept.   8) Teachers  guided  a  class  debriefing  on   the  experience  and  stimulated  meta-­‐cognitive  reflections  on   how  to  produce  good  narrative  texts.   9) At  the  end  of  the  school  year,  teachers  used  an  additional  video  stimulus  taken  from  World  of  Comedy   by  Harold  Lloyd  to  collect  a  second  round  of  individual  texts.  This  video  was  different  from  the  first   one,   in  order   to  assess   the   transferability  of   skills   across   tasks,   although   the   two   inputs  had   some   similar   characteristics   (e.g.,   number   of   main   characters,   plot   complexity,   duration)   and   were   administered  in  identical  conditions.   In   terms  of  White  &  Arndt's   (1991)  process   approach   to  writing,   Steps  2   and  3   correspond   to   the   stages  of  Generating  ideas  and  Focusing;  Step  4  corresponds  to  Text  Structuring,  and  step  5  to  Drafting.  The   processes  of  Evaluating  and  Re-­‐viewing  were  covered  in  Steps  6  and  7.   The   activities   above   were   just   a   part   of   the   experimental   classes’   syllabus,   accounting   for   approximately   two  hours  a  week  of   class   time  over   the  entire  school  year.  The  project   involved  all   classes,   from   Grades   1   to   5.   Obviously,   activities   for   the   first   two   grades   were   simplified   and   involved  much   less   writing,   but   they   still   followed   the   general   orientation   of   engaging   with   different   phases   of   the   writing   process  one  by  one.  Given  that   the  data  discussed   in   the   following  pages  come  from  Grades  3-­‐5,  no   further   details   on   the   first   two   grade   levels   are   offered  here:   for  more   information   about   the   teaching  phases   and   2   All   texts   are   translated   from   Italian.   Originals   are   available   on   the   author’s   website:   http://www.gabrielepallotti.it.   Concerning   translations,   when   the  morpho-­‐syntactic   and   orthographic  mistakes  made   by   the   pupils   did   not   have   any   straightforward  equivalent  in  English,  similar  types  of  errors  were  introduced  in  order  to  give  a  better  sense  of  the  overall   writing  quality.   PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 6   methods  and  the  variations  introduced  in  different  classes  and  age  levels,  see  Pallotti  (2017a,  2017b),  Pallotti   &  Rosi  (2017)  and  the  project  website  (interlingua.comune.re.it).     3.4.  Data  collection To   evaluate   the   effectiveness   of   the   training   program,   individual   texts   were   collected   in   seven   experimental  and  seven  control  classes,  Grades  3-­‐5  (ages  8-­‐11),  at  the  beginning  and  end  of  the  school  year.   Most   experimental   classes   had   been   participating   in   the   project   since   Grade   1,   so   that   at   the   time   of   data   collection  they  had  already  been  performing  these  types  of  activities   for   two,   three,  or   four  years.  Pupils   in   control   classes   at   the   beginning   and   end   of   the   school   year   wrote   their   texts   under   exactly   the   same   conditions  as  their  peers   in  the  experimental  classes  (i.e.,  They  watched  the  two  video  clips  by  Chaplin  and   Lloyd  twice)  and  were  given  the  same  generic  assignment  (i.e.,  Describe  the  video  to  a  teacher  who  has  not   seen  it),  with  no  directions  about  text  planning  and  writing  and  no  time  limit  to  accomplish  the  task.  Control   classes  followed  a  rather  regular  language  education  curriculum  (for  Italy),  in  which  writing  activities  mainly   consisted  in  composing  short  essays  or  narrative  texts,  training  in  orthography  and  grammar,  or  answering   comprehension  questions  after  reading.  Little  or  no  attention  was  devoted  to  individual  phases  of  the  writing   process,  no  artefacts  were  used  as  semiotic  mediators  for  higher-­‐order  cognitive  functions  like  text  planning   and  revision,  and  most  writing  activities  had  no  real  communicative  intent  but  were  conceived  as  exercises  to   be  assessed  by  teachers,  who  were  the  only  feedback  providers.   Although  great  care  was  taken  to  ensure  that  control  classes  paralleled  experimental  classes  as  much   as  possible,  the  correspondence  was  not  perfect,  as  is  often  the  case  with  quasi-­‐experimental  studies.  Control   classes  came  from  the  same  schools  or  areas  of  the  city  as  the  experimental  ones  and  they  had  a  comparable   number  of  pupils  per   class.  However,   their  demographic   composition  differed   from  multilingual   classes,   as   shown   in   Table   1.   Experimental   classes   had  more  multilingual   children,   over   one   third   of  whom  were   not   born  in  Italy  and  had  spent  an  average  of  just  two  years  in  the  country.  Control  classes  had  fewer  multilingual   pupils,  less  than  one  fifth  of  them  were  not  born  in  Italy  and  the  average  length  of  residence  in  Italy  for  these   children  was  longer  than  in  experimental  classes.     Table 1 Group composition Control Experimental Sample size 118 106 % multilingual pupils in entire class 24% 36% % multilingual pupils not born in Italy among all multilingual pupils 18% 38% Mean period of residency in Italy for multilingual pupils born abroad 3.6 yrs 2.1 yrs   A   second   difference   concerns   the   scores   at   the   beginning   of   the   school   year,   which,   for   several   dimensions   discussed   in   this   article,   were   higher   in   experimental   classes.   While   this   may   be   seen   as   unfortunate   from   a   strictly   scientific   point   of   view,   from   a   pedagogical   perspective   this   difference   is   quite   positive,   since   it   seems   to   indicate   that   classes   taking   part   in   the   experimental   program   from   first   grade   gradually  accrue  an  advantage  that  sets  them  apart  from  those  not  involved  in  the  program.  This  assumption   is   substantiated  by   the   fact   that,   for  most  measurements,   the   initial   gap  between  experimental  and  control   classes  was  lower  in  the  third  grade  and  higher  in  the  fourth  and  fifth  grades—that  is,  the  gap  grew  larger  in   parallel   to  a   longer  participation   in  the  project.  Furthermore,   the  experimental  group’s  higher  scores  at   the   beginning   of   the   school   year   cannot   be   ascribed   to   the   fact   that   these   pupils   came   from   privileged   socio-­‐ cultural  environments;  on  the  contrary,  most  experimental  classes  were  located  in  somewhat  disadvantaged   areas  of  the  city.   Given  this  difference  in  initial  scores,  this  article  focuses  on  final  results  only,  as  a  way  of  comparing   the  program’s  effects  accrued  after   three,   four,  or   five  years.   In  other  words,   it   seems  more  appropriate   to   assess   the   cumulative  effects  of   the  experimental  program  (which  began   in   first   grade   for  all   experimental   classes),   rather   than  discussing   the   gains  obtained  by   the  pupils   from   the  beginning   to   the   end  of   just   one   school  year.       AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 7   3.5.  Data  analysis   Handwritten  texts  were  first  transcribed  and  assigned  arbitrary  codes  that  did  not  allow  raters  and   coders  to  know  whether  they  came  from  experimental  or  control  classes,  or  whether  they  were  the  work  of   monolingual   or  multilingual   students,   to   avoid   any   possible   bias   in   the   procedures   requiring   a   qualitative   evaluation.  Pupils  diagnosed  with  cognitive  or  linguistic  impairment,  or  who  did  not  attend  the  initial  or  final   data   collection   sessions,   were   excluded   from   the   analysis,   which   was   thus   based   on   106   participants   in   experimental  classes  and  118  in  control  classes.  Finally,  scripts  were  examined  both  holistically  and  by  means   of  analytic  measurements. Holistic   examination   was   carried   out   using   scales   to   rate   overall   communicative   effectiveness.   In   particular,  functional  adequacy  was  rated  with  three  scales  slightly  adapted  from  Kuiken  and  Vedder  (2017),   while  a  scale  taken  from  the  Common  European  Framework  (Council  of  Europe,  2001)  was  used  for  assessing   text  coherence  and  cohesion.  The  scales  are  reported  in  Appendix  A.  In  order  to  assess  inter-­‐rater  reliability,   10%  of  the  scripts  were  rated  on  the  four  scales  by  two  independent  researchers,  who  showed  a  substantial   level  of  agreement:   the  average   Intraclass  correlation  coefficient   (two-­‐way  mixed  effects  model)  was   .85. A   series   of   features   was   then   considered   for   the   analytical   measurement   of   text   properties,   covering   the   dimensions  outlined  below.       3.5.1.  Content   This  dimension  was  assessed  by  looking  at  text  length,  measured  by  the  number  of  words  produced   and  by  counting  the  number  of  primary  and  secondary  idea  units.  Similar  to  other  studies  (e.g.,  Boscolo,  Arfé,   &   Quarisa,   2007),   these  were   agreed   upon   by   the  whole   research   team,   who   produced   a   list   of   five  main   information   units,   representing   the   story’s   key   sequences,   and   43   secondary   units,   representing   specific   events  in  the  story  plot.     3.5.2.  Text  organization   This  dimension  refers  to  the  pupils’  ability  to  effectively  divide  and  organize  their  text  into  thematic   and   syntactic   blocks.   It   is   evaluated   firstly   by   counting   the   number   of   paragraphs,   and   their   correlated   measure,  the  number  of  words  per  paragraph.  Paragraphing  is  normally  not  taught  in  Italian  schools,  so  that   pupils  often  end  up  writing  texts  in  just  one  single  long  block.  Our  pedagogic  intervention  explicitly  focused   on  this  skill,  because  it  was  considered  closely  related  to  planning  the  text  in  several  thematic  units.   At   a   different   level   of   textual   organization,   the   comma   serves   to   divide   syntactic   units,   and   its   appropriate   use   is   rarely   taught   in   Italian   schools,   where   it   is   often   simplistically   explained   as   a   way   of   marking  pauses  and  intonation  or  to  separate  items  in  a  list.  Pupils  are  introduced  to  this  punctuation  mark   between  Grades   2   and  3   and   they   soon   start   using   it,   initially   in   a   few  prototypical   contexts,   such   as   lists.   Gradually,   its   usage   is   expanded   to   express  more   sophisticated  meanings,   like   differentiating   levels   of   text   organization  or  signalling  discourse  structure,  but   this  comes  with  a  higher  risk  of  making  mistakes,  which   persist   throughout   middle   and   high   school   with   pupils   rarely   being   completely   aware   of   their   nature   (Solarino,   2009).   Thus,   a   higher   number   of   commas  may  be   taken   as   an   indicator   of   pupils’  willingness   to   experiment   with   this   important   but   challenging   and   often   misunderstood   punctuation   mark.   However,   a   greater  use  of  the  comma  should  not  be  automatically  interpreted  as  an  indicator  of  better  text  quality.  This  is   why   the   proportion   of   inappropriate   commas   to   all   commas   used   was   measured,   too.   Omissions   were   intentionally   not   counted,   since   establishing   when   a   comma   is   “missing”   is   often   a   matter   of   discretion.   Another   indicator   of   a   good   command   of   punctuation   is   the   ratio   of   inappropriate   uses   of   the   period.   However,  the  number  of  periods  per  100  words  was  not  considered  to  be  a  clearly  interpretable  measure:  on   the  one  hand,  it  may  signal  a  desirable  feature,  that  is  a  higher  degree  of  text  segmentation  and  avoidance  of   run-­‐on  sentences;  on  the  other,  it  may  imply  a  very  paratactic  style,  with  lots  of  simple,  telegraphic  sentences,   which  do  not  necessarily  contribute  to  text  quality.   3.5.3.  Reference  to  entities   Managing   reference   to   entities   in   discourse   is   a   notoriously   difficult   task   for   young   learners,   and   sometimes   for   adult   speakers   and   writers   as   well   (Giuliano,   2007;   Hendriks,   1998;   Hickmann,   1995).   An   entity  can  be  introduced,  upon  its  first  mention,  and  then  maintained  in  the  following  clauses.  This  referential   chain   can   be   interrupted   with   the   introduction   of   new   entities,   but   the   original   entity   can   be   later   on   reintroduced  in  the  text.  In  the  following  Fictive  Text  (1),  introductions  are  indicated  by  (I),  maintenances  by   PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 8   (M),  and  reintroductions  by  (R).  Coding  and  analysis  were  conducted  on  animate  entities  only,  as  they  are  a   key  element  in  plot  progression  and  thus  play  a  central  role  in  narrative  texts.     Fictive Text (1) There was a boy(I) who went to the park. He(M) was very shy and Ø(M) had very few friends. An old man(I) came to the park and Ø(M) took out a worn-out kite. He(M) looked a bit lonesome, too. The boy(R) asked him(M) if they(M) could play together   Normally,  entities  are  introduced  by  lexical  noun  phrases  with  indefinite  determiners  (a  boy,  an  old   man),  they  are  maintained  by  pronouns  or  zero  anaphora  (he,  Ø,  him)  and  are  re-­‐introduced  by  lexical  noun   phrases  with  definite  determiners  (the  boy).  A  common  feature  of  young  children’s  speech  and  writing  is  that   entities  are  often  introduced  with  definite  determiners,  so  that,  in  the  example  above,  the  first  sentence  could   have  been  something  like,  The  boy  went  to  the  park.  These  introductions  may  be  called  inappropriate,  as  they   indicate   that   their   author   improperly   assumes   the   referent   is   accessible   to   the   hearer/reader.   However,   although   they  may   sound   childish   or   slightly   egocentric,   they   do   not   compromise   communication   and   the   meaning   remains   clear.   In   this   article,   however,   these   phenomena   will   not   be   discussed   (for   further   discussion  on  this  aspect,  see  Pallotti,  Borghetti,  &  Rosi,  in  preparation). What   will   be   focused   on   here   are   ambiguous   references   that   prevent   the   receiver   from   correctly   interpreting  their  meaning.  If  the  text  above  had  begun,  He  went  to  the  park,  this  would  have  been  counted  as   an  ambiguous  introduction.  Likewise,  if  the  last  sentence  had  been,  He  asked  him  if  they  could  play  together,  it   would   have   been   impossible   to   establish   who   was   asking   whom.   Only   by   knowing   the   intended  meaning   (which  is  possible,  for  instance,  when  pupils  are  retelling  a  film  or  a  picture  story)  can  one  score  the  first  he  as   an   ambiguous   reintroduction   and   the   immediately   following   him   in   this   sentence   as   an   ambiguous   maintenance.     3.5.4.  Verb  tense  cohesion   Another  common  feature  of  young  children’s   language   is   the  frequent  shifting  of  verb  tenses   in  the   same  narrative  (Kersten,  2009),  as  in  sentences  like:  Then  he  starts  skating  and  she  went  to  bed.  A  verb  shift  is   defined   as   a   change   of   verb   tense   that   is   not   justified   for   textual   reasons,   such   as   the   move   from   the   description  of  background  states  to  the  narration  of  events  in  the  plot,  or  from  indirect  to  direct  speech.  The   operational  definition  of  all  these  categories  was  set  out  in  a  detailed  coding  manual,  which  provided  specific   examples  for  each  phenomenon,  including  possible  counterexamples  and  borderline  cases.   As   regards   the   statistical   procedures   employed   in   the   analysis,   given   that   the   study   was   a   quasi-­‐ experimental   investigation   of   intact   classes   and   that   initial   levels   were   not   homogenous   for   several   dimensions,  Welch  t-­‐tests  for  independent  samples  were  employed  to  compare  the  averages  of  experimental   and  control  classes  at  the  end  of  the  school  year.  According  to  Fagerland  and  Sandvik  (2009),  this  test  is  best   suited   for  samples   like  ours,  which  consist   in  about  100  cases,  normally  distributed  or  moderately  skewed,   with  similar  but  not  identical  variances.  For  holistic  assessment  data,  whose  distributions  were  substantially   normal,  t-­‐tests  without  Welch  correction  were  used,  since  they  are  more  appropriate  for  Likert-­‐like  data,  with   many  tied  ranks  (de  Winter  &  Dodou,  2010).     4.  Results   This   section   reports   on   the  main   results   regarding   program   effectiveness   at   the   end   of   the   school   year.  Tables  include  the  mean  and  standard  deviation,  the  95%  confidence  interval  for  the  estimation  of  the   difference   between   the  means   of   experimental   and   control   classes,   Cohen’s  d   effect   size,   calculated   as   the   ratio  between  the  difference  between  means  and  the  standard  deviation  of  the  control  group,  and  the  t-­‐test’s   two-­‐tailed  p  values.  Results  will  be  reported  firstly  for  entire  classes,  then  for  monolingual  and  multilingual   sub-­‐groups.         4.1.  Holistic  assessment At  the  end  of  the  school  year,  text  assessment  through  rating  scales  showed  very  positive  results  for   the  experimental  classes  as  a  whole.  Their  compositions  were  perceived  as  being  richer  in  terms  of  content   and  more   coherent   and   cohesive,  both   considering  Kuiken  and  Vedder’s   (2017)   scale   and   for   the  Common   AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 9   European  Framework  scale  (Council  of  Europe,  2001)  with  rather  large  effect  sizes  of  between  0.61  and  0.78.   They  were  also  more  comprehensible,  with  a  smaller  but  still  statistically  significant  difference  (d  =  0.36). This   trend   is   also   confirmed   for   multilingual   pupils,   who   wrote   more   complete,   coherent,   and   cohesive  texts  than  their  control  peers,  although  differences  were  less  marked  (d  between  0.39  and  0.53)  and   even   smaller   when   compared   to   monolingual   students   in   control   classes   (d   between   0.02   and   0.18).   Multilingual   pupils   in   experimental   classes   scored   slightly   lower   (M   =   3.24)   than   multilingual   and   monolingual   peers   (M   =   3.41)   in   control   classes   only  with   regards   to   text   comprehensibility   however   the   difference   is   not   significant.   Monolingual   pupils   in   experimental   classes   largely   outperformed   their   monolingual  peers  in  control  classes  in  all  dimensions.    These  overall  results  are  shown  in  Table  2.     Table 2 Overview of results from holistic assessment Control Group Mean (SD) Experimental Group Mean (SD) 95% CI of the difference between means Cohen’s d p Content (1-6) 3.39 (1.13) 4.09 (1.06) 0.39 - 0.99 0.61 < .0001 Monolingual pupils 3.46 (1.19) 4.32 (1.07) 0.50 - 1.22 0.72 < .0001 Multilingual pupils 3.19 (0.92) 3.68 (1.08) −0.99 - 0.01 0.53 (ns) Comprehensibility (1-6) 3.53 (0.93) 3.86 (1.01) 0.07 - 0.60 0.36 < .0001 Monolingual pupils 3.56 (0.95) 4.21 (0.87) 0.36 - 0.94 0.68 < .0001 Multilingual pupils 3.41 (0.89) 3.24 (0.98) −0.30 - 0.64 -0.18 ns Coherence-Cohesion KV (1-6) 2.89 (0.81) 3.47 (1.04) 0.33 - 0.82 0.72 < .0001 Monolingual pupils 2.95 (0.82) 3.74 (0.92) 0.50 - 1.07 0.96 < .0001 Multilingual pupils 2.65 (0.73) 2.97 (1.01) −0.74 - 0.13 0.42 ns Coherence-Cohesion CEF (1-8) 3.25 (1.32) 4.27 (1.40) 0.66 - 1.39 0.78 < .0001 Monolingual pupils 3.33 (1.31) 4.71 (1.25) 0.97 - 1.79 1.05 < .0001 Multilingual pupils 2.96 (1.34) 3.49 (1.39) −1.21 - 0.17 0.39 ns Note. KV = rating scale by Kuiken and Vedder (2017); CEF = rating scale of the Common European Framework (Council of Europe, 2001, 2017).   4.2.  Content   We  now  turn  to  analytical  measures  of  text  quality.  The  texts  produced  in  experimental  classes  are   much  longer  than  those  of  control  classes,  with  an  average  of  243.84  words  versus  177.93;  they  also  contain   more  secondary  information  units  (22.56  vs.  17.81)  and  thus  provide  a  more  detailed  and  rich  content,  with   considerable   effect   sizes   (0.83   and   0.78,   respectively).   The   difference   is   smaller,   but   still   statistically   significant,   as   regards  main   idea   units   (4.79   vs   4.60),   since  most   pupils,   in   both   experimental   and   control   classes,  were  able  to  mention  all  or  almost  all  of  them.   The   trends   identified   at   the   entire-­‐class   level   are   also   confirmed   by   multilingual   pupils   in   experimental   classes,  who  write   longer   texts   than  multilingual   students   in   the   control   group.   Interestingly,   their  productions  are  just  slightly  shorter  than  those  of  their  monolingual  peers  in  experimental  classes,  and   considerably  longer  than  the  texts  written  by  monolingual  pupils  in  the  control  group.   This  greater  text  length  reflects  the  number  of  primary  and  secondary  information  units.  The  number   of  units   is  higher  among  the  multilingual  students   in  the  experimental  group  than   in  both  monolingual  and   multilingual  controls.  Similar  to  the  results  discussed  in  the  previous  section,  the  pupils  in  the  experimental   groups   who   demonstrate   the   greatest   advantage   over   their   counterparts   in   the   control   groups   are   the   monolingual  students  (see  Table  3).           PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 10   Table 3 Content (analytic measures) Control Group Mean (SD) Experimental Group Mean (SD) 95% CI of the difference between means Cohen’s d p Words 177.93 (79.12) 243.84 (96.67) 41.82 - 90.00 0.83 < .0001 Monolingual pupils 170.09 (69.04) 258.95 (97.00) 61.03 - 116.70 1.29 < .0001 Multilingual pupils 203.19 (102.83) 216.89 (96.76) −64.60 - 37.19 0.13 ns Main idea units (0-5) 4.60 (0.65) 4.79 (0.41) 0.04 - 0.34 0.29 < .05 Monolingual pupils 4.61 (0.64) 4.82 (0.46) 0.034 - 0.38 0.33 < .05 Multilingual pupils 4.56 (0.70) 4.73 (0.51) −0.49 - 0.14 0.25 ns Secondary idea units (0-43) 17.81 (6.12) 22.56 (6.13) 3.06 - 6.45 0.78 < .0001 Monolingual pupils 17.41 (5.81) 23.30 (6.34) 3.91 - 7.86 1.01 < .0001 Multilingual pupils 19.07(7.00) 21.24 (6.68) −5.65 - 1.32 0.31 ns 4.3.  Text  organization   Students   in   experimental   classes   started   new   paragraphs   more   frequently   than   those   in   control   classes,  with  an  average  of  5.32  paragraphs  per  text  versus  2.39.  As  a  result,  the  mean  number  of  words  per   paragraph   was   lower:   74.82   as   opposed   to   116.89.   These   differences   are   statistically   significant   with   medium-­‐to-­‐high  effect  sizes  (see  Table  3).     Pupils   in   experimental   classes   used   the   comma   more   frequently   (3.63/100   words   vs.   2.34/100   words),  while  their  use  of  periods   is  virtually   identical   to  that  of  control  classes  (3.50  vs.  3.63/100  words),   meaning  that  their  sentences  were  about  the  same  length. Inappropriate  commas,  e.g.  those  separating  a  verb  from  its  necessary  arguments,  were  also  scored.   In  spite  of  their  more  frequent  use  of  the  comma,  which  implies  taking  more  risks  (and  probably  committing   fewer  errors  of  omission,  which  however  were  not   scored  because  of   the   limited   reliability  of   this  analytic   category),  pupils  in  experimental  classes  made  slightly  fewer  errors  than  those  of  control  classes  (13.38%  vs.   15.73%  of  all  commas),  and  significantly  fewer  errors  regarding  the  period  (1.44%  vs.  5.14%). Multilingual   pupils,   once   again,   have   a   substantially   similar   profile   to   that   of   the   monolingual   students   in   each   group.   Those   in   experimental   classes   start   about   twice   as   many   new   paragraphs   as   monolingual  and  multilingual  pupils   in  control  classes.  As  a  result,   their  paragraphs  are  on  average  shorter   (70.58  words  per  paragraph).  Like  their  monolingual  peers  in  the  experimental  group,  they  use  commas  more   frequently   than  multilingual  and  monolingual  students   in  control  classes:  2.63  commas/100  words  vs.  2.03   and  2.44,  respectively.  These  commas  are  on  average  more  correct  than  those  produced  by  the  monolingual   control   pupils   (15.18%   vs.   17.35%   of   incorrect   commas),   but   slightly   less   so   than   those   used   by   the   multilingual   control   pupils   (9.67%   of   incorrect   commas),   perhaps   because   the   latter   subgroup   used   this   punctuation  mark  the  least.     Table  4  below  reports  an  overview  of  results  for  each  feature.  Also,  texts  in  Extract  (1)  exemplify  two   different  phenomena  related  to  paragraphing  and  use  of  punctuation  in  a  third-­‐grade  pupil  belonging  to  the   experimental  group.3                   3  All   texts  are   translated   from  Italian.  Originals  are  available  on   the  author’s  website:  www.gabrielepallotti.it.  Concerning   translations,   when   the   morpho-­‐syntactic   and   orthographic   mistakes   made   by   the   pupils   did   not   have   any   straightforward   equivalent   in   English,   similar  types  of  errors  were  introduced  in  order  to  give  a  better  sense  of  the  overall  writing  quality.   AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 11   Table 4 Text organization Control Group Mean (SD) Experimental Group Mean (SD) 95% CI of the difference between means Cohen’s d p Paragraphs 2.39 (2.00) 5.32 (3.28) −3.68 - −2.19 1.47 < .0001 Monolingual pupils 2.36 (1.99) 5.70 (3.41) −2.41 - −4.27 1.68 < .0001 Multilingual pupils 2.48 (2.06) 4.65 (3.10) 0.87 - 3.46 1.05 < 0.005 Words/paragraph 116.89 (81.57) 74.82 (80.21) −20.36 - −63.78 −0.52 < .0001 Monolingual pupils 115.96 (83.24) 77.20 (89.64) −10.68 - −66.85 −0.47 < .001 Multilingual pupils 119.90 (77.37) 70.58 (61.60) 13.12 - 85.52 −0.64 < 0.01 Commas/100 words 2.34 (2.22) 3.63 (2.31) −1.93 - −0.65 0.58 < .005 Monolingual pupils 2.44 (2.25) 4.19 (2.55) 0.97 - 2.53 0.78 < .0001 Multilingual pupils 2.03 (2.15) 2.63 (2.12) −1.68 - 0.48 0.28 ns % inappropriate commas 15.73 (22.77) 13.38 (16.03) −3.54 - 8.13 −0.10 ns Monolingual pupils 17.35 (24.39) 12.55 (16.14) −0.02 - 0.12 −0.20 ns Multilingual pupils 9.67 (14.22) 15.18 (16.51) −14.66 - 3.59 0.38 ns % inappropriate periods 5.14 (15.64) 1.44 (4.87 0.60 - 6.78 −0.24 < .05 Monolingual pupils 5.88 (17.04) 0.96 (2.83) −1.17 - −9.69 −0.29 < .05 Multilingual pupils 2.60 (9.26) 2.28 (6.79) −4.03 - 4.69 −3.48 ns     Extract (1) Text organization at the beginning and at the end of the school year, pupil 3OL60 3OL60, October First Charlot shows a ticket to a man after that the man welcoms him Charlot then Charlot walks around with a man after that Charlot takes a lady by the hand and takes her into the warehouse after that they put on roller skates then they start skating after that Charlot blindfolds himself and starts to rolerskate then it looks like Charlot is falling down from the upper floor after that Charlot shows a bed to the lady then Charlot moves around a bit on his rollerskates then Charlot closes the girl in the warehouse after that 3 gentlemen look for Charlo who in the meantime is in a totally different place 3OL60, May At first Harry gets on a steam train then a girl climbs in with a hat and a Pomeranian dog. The girl droped the Pomeranian dog and they left without him but Harry took it with a cane and gave it to its owner named Giulia, she was very happy to get back her Pomeranian dog, and told him thank you very much. At some point they entered a tunnel and everyone started looking for the dog. Harry found it and gave it to Giulia once again, s/he was really really happy to have it back, and then Giulia covered it with a blanket of hers/his.   While  both  paraphrasing  and  punctuation  marks  are  absent   in   the   first   text,   at   the  end  of   the   school  year  the  child  organizes  contents  in  thematic  blocks  and  experiments  with  the  use  of  punctuation,   even  though  her  command  of  commas  is  not  fully  developed  (see  the  last  two  instances).       4.4.  Reference  to  entities   At  the  end  of  the  school  year,  the  texts  written  in  the  experimental  classes  were  also  clearer  than  the   texts  written  in  the  control  classes  in  terms  of  reference  to  entities.  In  fact,  the  introduction,  maintenance,  and   reintroduction   of   referents  were   less   ambiguous,   that   is,   they   allowed   the   reader   to   correctly   identify   the   entity  to  which  they  refer  more  often.  This  difference  was  not  very  substantial  as  regards  introductions  and   maintenances,  but  reached  statistical  significance  for  reintroductions  (p  <  0.05). PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 12   As  for  multilingual  pupils,  the  picture  is  slightly  more  complex.  Their  reintroductions  were  generally   less   ambiguous   than   those   of   both  monolingual   and  multilingual   students   in   control   classes,   with   a   fairly   noticeable,   though  not   statistically   significant,   difference   (6.37%  vs.   8.42%  and  9.36%,   respectively).   Their   introductions  were  less  ambiguous  than  those  of  monolingual  pupils  in  the  control  group  (1.79%  vs.  2.39%),   but  significantly  more  so  than  those  made  by  multilingual  peers  in  the  control  classes,  who  did  not  produce  a   single   ambiguous   introduction.   A   similar   situation   can   be   seen   for   maintenances:   here,   too,   multilingual   pupils   in  experimental   classes  produced   fewer  ambiguities   than  monolingual   students   in   the  control  group   (5.66%  vs.  5.84%),  but  slightly  more   than  multilingual  peers   (5.50%).  Especially   in   this   last  case,  however,   differences   are  quite   small   and  never   reach   statistical   significance   (Table  5).   Some  examples  of   ambiguous   introductions,   maintenances,   and   reintroductions   taken   from   both   experimental   and   control   classes   are   shown  in  Extract  (2),  (3),  and  (4).     Table 5 Reference to entities Control Group Mean (SD) Experimental Group Mean (SD) 95% CI of the difference between mean Cohen’s d p % Ambiguous introductions Monolingual pupils Multilingual pupils 1.83 2.39 0.00 (6.49) (7.35) 1.27 (3.47) 0.98 (3.47) 1.79 (4.69) −0.86 −0.36 0.23 - - - 1.99 3.19 3.33 −0.09 −0.19 n/a ns ns <0.05 % Ambiguous maintenances Monolingual pupils Multilingual pupils 5.76 5.84 5.50 (9.22) (9.33) (.04) 5.34 (5.70) 5.16 (5.64) 5.66 (6.80) −1.70 −1.80 −4.37 - - - 2.47 3.03 4.10 −0.05 −0.07 0.02 ns ns ns % Ambiguous reintroductions Monolingual pupils Multilingual pupils 8.64 8.42 9.36 (9.70) (9.31) (11.05) 5.68 (6.85) 5.29 (6.89) 6.37 (8.04) −0.71 −0.56 −2.01 - - - −5.28 −5.75 −8.05 −0.31 −0.34 −0.27 < .05 < .05 ns     Extract (2). Example of ambiguous references to entities Ambiguous introduction 4VM19_October [...] when he came down with the elevator the men [new entity] hide from Charlot. Extract (3). Example of ambiguous references to entities Ambiguous maintenance 3EP05_May [...] then the ledy looked at the clock and tooked the bag where the dog was then Ø [= she/he/it?] begun to bark and the ledy opened the bag and faund the dog but not her handkerchiefs [...] Extract (4). Example of ambiguous references to entities Ambiguous reintroduction 5ARR03_May [...] Harry saw the ticket controller arrive so he took his suitcase, which was the woman’s he began to take out all the clothes, then he put the dog in and put it back when the ticket controller passed Ø [= he/it?] started barking so the ticket controller became suspcious, Harry, began to bark to create confusion. [...]       AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 13   4.5.  Verb  tense  cohesion   The  one  aspect  in  which  experimental  classes  performed  more  poorly  than  the  controls  is  verb  tense   cohesion,  where  unjustified  tense  changes,  here  labelled  “tense  shifts,”  were  more  common:  2.43/100  words   vs.  1.58/100  words.  Extract  (5)  reports  an  example  from  the  experimental  group.       Extract (5). Tense shifts in a fourth grader, experimental group 4GC43_May Harry a man has to take a train and he saw a lady and lets her pass first. The lady start talking to the ticket controller and Harry came in quickly. The lady's little dog has jumped off the train and she wanted to get it back. Harry picked up a can and catches the little dog and hid it in his jacket. [...]   It  is  worth  noting  that  the  text  this  extract  comes  from  is  of  average  length  (248  words),  is  quite  rich   in  terms  of  primary  and  secondary  information  units  (5  and  29,  respectively),  and  presents  a  good  quantity  of   cohesive  devices  (28  connectives).  This  means  that  verb  tense  cohesion  may  still  pose  problems,  even  when   overall  text  quality  is  satisfactory. The  difference  seen  between  experimental  and  control  classes  as  a  whole  remains  basically  the  same   when  multilingual   and  monolingual   pupils   are   analyzed   separately:   multilingual   students   in   experimental   classes   produce   a   higher   proportion   of   tense   shifts   (3.91/100   words)   than   multilingual   (2.63)   and   monolingual   (1.26)  peers   in   the   control   group;  monolingual  pupils   in   experimental   classes,  with   their  1.59   shifts   per   100   words,   perform   better   than   multilingual   and   slightly   worse   than   monolingual   controls.   It   should   be   noted   that,   in   both   experimental   and   control   classes,   this   phenomenon   is   more   apparent   in   multilingual  than  in  monolingual  pupils.  Tense  shifts  are,  in  fact,  more  closely  related  to  linguistic  competence   in  a  narrow  sense,  that  is,  to  grammatical  control  rather  than  to  the  more  communicative  aspects  of  writing   ability.  This  finding  resonates  with  a  number  of  studies  which  have  reported  that  LX  writers  have  less  control   over  verb  forms  compared  to  L1  peers,  both  during  compulsory  education  (Cameron  &  Besser,  2004)  and  at   the  university  (Hinkel,  2004).   5.  Discussion  and  conclusion   As  discussed  above,  this  program  was  proposed  to  the  whole  class,  without  dividing  monolingual  and   multilingual  pupils.  The  underlying  principle  is  that  everybody  benefits  from  good  quality  language  education   and   that   promoting   peer-­‐to-­‐peer   interaction   through   group   work   gives   better   results   than   having   pupils   follow  separate  tracks.   Although   there   was   no   separation   of   learners   in   terms   of   teaching,   it   is   legitimate   to   discuss   the   program’s  effects  on  the  two  sub-­‐groups  of  monolingual  and  multilingual  students.  Not  many  results  at  this   level   are   statistically   significant,   both  because   the   sub-­‐samples  were  not   very   large   and  because   in   several   cases   differences   were   not   very   pronounced.   Thus,   this   brief   discussion   will   not   aim   at   reaching   broad   generalizations,   rather   it  will   attempt   to   stimulate   reflection   on   the   relationships   between  multilingualism   and  language  education  in  today’s  Italian  classrooms,  which  are  possibly  applicable  to  other  contexts  as  well.   The  study's  main  findings,  based  on  the  research  questions  formulated  in  section  3.1,  can  be  summarized  as   follows.   RQ1.   In   a   primary   school   context,  what   are   the   effects   on  whole   classes   participating   in   a   long-­‐term   pedagogical   intervention   based   on   the   process   approach   to   writing,   formative   assessment,   and   cooperative   learning?  The  effects  on  whole  classes  are  largely  positive.  Their  pupils  outperform  those  of  control  classes  on   all  dimensions  assessed  with  both  analytic  and  holistic  measures,  the  sole  exception  being  verb  tense  shifts.   This   result   is   thus   in   line   with   previous   research   (reviewed   in   section   2.2)   showing   the   effectiveness   of   instructional   practices   like   teaching   students   strategies   for   planning,   drafting,   and   revising   their   texts,   promoting  peer  collaboration,  and  providing  constructive  and  specific  feedback.     RQ2.   If   there  are  any  effects,   do   they  differ   for  monolingual  and  multilingual  pupils?   In  experimental   classes,  monolingual  children  achieve  better  results  than  their  multilingual  peers  with  regard  to  all  aspects  of   writing   with   one   exception,   the   word   per   paragraph   ratio.   It   should   be   noted,   though,   that   experimental   classes  hosted  a  higher  number  of  multilingual  students  who  were  not  born  in  Italy  and  who  had  resided  in   the  country  for  a  shorter  period  of  time;  hence,  their  relatively  poorer  performance  might  be  due  to  language   learning   per   se,   as  many   of   these   children  were   true   additional   language   learners,   whose   Italian  was   still   developing.   PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 14   Despite  this  initial  disadvantage,  the  program's  effects  were  equally  beneficial  for  both  monolingual   and  multilingual  students.  In  particular,  multilingual  children  in  the  experimental  group  outperformed  both   monolingual  and  multilingual  students  in  control  classes  on  the  vast  majority  of  dimensions,  with  just  a  few   exceptions.  One  of  these  is  tense  shifts,  which  were  consistently  more  frequent  in  experimental  classes  (with   the   sole   exception   of   monolingual   pupils   in   experimental   classes   performing   better   than   multilingual   controls);   this   shows   that   tense   shifts  may   be  more   directly   related   to   language   proficiency   than   to   other   dimensions.  Another  aspect   that  may  be   related   to   language  proficiency   is   text   comprehensibility,  which   is   lower   in   experimental   multilingual   pupils   when   compared   to   both   monolingual   and   multilingual   control   students.  Finally,  multilingual  control  children  outperform  experimental  (both  monolingual  and  multilingual)   peers  as  regards  appropriate  use  of  the  comma  and  ambiguous  introductions.   RQ3.  What  are  the  differences,  as  regards  writing  skills,  between  monolingual  and  multilingual  pupils  in  control   classes,   that   is,  with   no   experimental   treatment?   In   control   classes,  multilingual   and  monolingual   children’s   performances  seem  to  be   rather  balanced.  Monolingual  pupils  achieve  consistently  better   scores   in  holistic   assessment  and,  as  regards  analytic  measures,  tense  shifts,  their  use  of  commas,  and  they  have  a  lower  rate  of   ambiguous  reintroductions.  Multilingual  students,  on  the  other  hand,  write  longer  texts  with  more  secondary   information   units,   with   fewer   problems   in   the   use   of   periods   and   commas,   and   with   fewer   ambiguous   introductions  and  maintenances.   The   tentative   conclusions   that   can   be   drawn   from   the   study   are   that,   at   least   in   this   sample,   participation   in   the  experimental  program  seems  to  have  had  a   larger   impact  on  writing  performance   than   monolingual/multilingual   status.  Multilingual   children   in   experimental   classes   contribute   as  much   as   their   monolingual  peers  to  the  good  results  obtained  by  the  class  as  a  whole,  and  outperform  both  monolingual  and   multilingual   pupils   in   control   classes   in   most   dimensions,   especially   those   related   to   broader   “cognitive-­‐ academic”   (Cummins,   1986)   aspects.   Linguistic   proficiency,   however,   does   play   a   role,   especially   for   those   who  have  had  a  shorter  immersion  in  the  new  language,  as  was  the  case  for  several  multilingual  children  in   experimental  classes.   In  the  context  investigated  here,  our  general  conclusion  is  that  students’  writing  skills  are  impacted   by  both  intrinsic  factors,  such  as  the  length  of  exposure  to  the  target  language,  and  extrinsic  factors,  such  as   the  type  of  pedagogical  intervention.  This  study  can  be  taken  as  evidence  that  a  good  language  curriculum  has   more  weight  in  a  student’s  academic  success  than  his  or  her  status  as  an  L1  or  LX  learner.         Acknowledgments     We  wish  to  acknowledge  the  contribution  of  Fabiana  Rosi,  who  was  involved  in  the  design  of  the  educational   intervention  and  in  the  data  collection,  as  well  as  the  contributions  of  Giulia  Borrelli,  Elisa  Caretta,  and  Elena   Russo,  who  carried  out  some  of  the  data  coding  and  analysis.         References     Andrade,  Maureen  Snow,  &  Evans,  Norman  W.    (2013).  Principles  and  practices  for  response  in  second   language  writing.  New  York,  New  York:  Routledge.   Boscolo,  Pietro,  Arfé,  Barbara,  &  Quarisa,  Mara  (2007).  Improving  the  quality  of  students’  academic  writing:   an  intervention  study.  Studies  in  Higher  Education,  32(4),  419–438.   http://dx.doi.org/10.1080/03075070701476092.   Brakel  Olson,  Vicki  L.  (1990).  The  revising  processes  of  sixth-­‐grade  writers  with  and  without  peer  feedback.   The  Journal  of  Educational  Research,  84(1),  22–29.   Cameron,  Lynne,  &  Besser,  Sharon  (2004).  Writing  in  English  as  an  additional  language  at  key  stage  2.   Research  report  RR586.  Leeds,  United  Kingdom:  Department  for  Education  and  Skills,  University  of   Leeds.   Cook,  Vivian  J.  (1999).  Going  beyond  the  native  speaker  in  language  teaching.  TESOL  Quarterly,  33(2),  185– 209.  http://dx.doi.org/10.2307/3587717.   AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 15   Council  of  Europe  (2001).  Common  European  framework  of  reference  for  languages:  learning,  teaching,   assessment.  Cambridge,  United  Kingdom:  Cambridge  University  Press.   Cumming,  Alister  (2001).  Learning  to  write  in  a  second  language:  Two  decades  of  research.  International   Journal  of  English  Studies,  1(2),  1–23.  http://dx.doi.org/10.6018/ijes.1.2.48331.   Cumming,  Alister,  Rebuffot,  Jacques,  &  Ledwell,  Monica  (1989).  Reading  and  summarizing  challenging  texts  in   first  and  second  languages.  Reading  and  Writing,  2,  201–219.  http://dx.doi.org10.1007/BF00377643   Cummins,  Jim  (1986).  Empowering  minority  students:  A  framework  for  intervention.  Harvard  Educational   Review,  56(1),  18–37.   Dewaele,  Jean-­‐Marc  (2017).  Why  the  dichotomy  ‘L1  versus  LX  user’  is  better  than  ‘native  versus  non-­‐native   speaker’.  Applied  Linguistics,  39(2),  236–240.  http://dx.doi.org10.1093/applin/amw055.   De  Winter,  Joost  C.  F.,  &  Dodou,  Dimitra  (2010).  Five-­‐point  Likert  items:  t  test  versus  Mann-­‐Whitney-­‐ Wilcoxon.  Practical  Assessment,  Research  &  Evaluation,  15(11),  1–12.   Ede,  Lisa  S.,  &  Andrea  C.  Lunsford  (1990).  Singular  texts/plural  authors:  perspectives  on  collaborative  writing.   Carbondale  and  Edwardsville,  Illinois:  Southern  Illinois  University  Press.   Englert,  Carol  Sue,  Mariage,  Troy  V.,  &  Dunsmore,  Kailonnie  (2006).  Tenets  of  sociocultural  theory  in  writing   instruction  research.  In  Charles  A.  MacArthur,  Steve  Graham,  and  Jill  Fitzgerald  (Eds.),  Handbook  of   writing  research  (pp.  208–221).  New  York,  New  York:  Guilford  Press.     Fagerland,  Morten  W.,  &  Sandvik,  Leiv  (2009).  Performance  of  five  two-­‐sample  location  tests  for  skewed   distributions  with  unequal  variances.  Contemporary  Clinical  Trials,  30(5),  490–496.   http://dx.doi.org10.1016/j.cct.2009.06.007.   Ferretti,  Ralph  P.,  Lewis,  William  E.,  &  Andrews-­‐Weckerly,  Scott  (2009).  Do  goals  affect  the  structure  of   students’  argumentative  writing  strategies?  Journal  of  Educational  Psychology,  101,  577–589.   http://dx.doi.org10.1037/a0014702   Fitzgerald,  Jill,  &  Teasley,  Alan  B.  (1986).  Effects  of  instruction  in  narrative  structure  on  children's  writing.   Journal  of  Educational  Psychology,  78(6),  424–432.  http://dx.doi.org/10.1037/0022-­‐0663.78.6.424.   Giuliano,  Patrizia  (2007).  Abilità  narrativa  ed  emarginazione  sociale.  Napoli,  Italia:  Liguori.   Graham,  Steve,  Harris,  Karen  R.,  &  Chambers,  Amber  B.  (2016).  Evidence-­‐based  practice  and  writing   instruction:  A  review  of  reviews.  In  Charles  A.  MacArthur,  Steve  Graham,  and  Jill  Fitzgerald  (Eds.),   Handbook  of  writing  research  (pp.  211–226).  New  York,  New  York:  Guilford  Press.     Graham,  Steve,  Harris,  Karen  R.,  &  Santangelo,  Tanya  (2015).  Research-­‐based  writing  practices  and  the   common  core:  Meta-­‐analysis  and  meta-­‐synthesis.  The  Elementary  School  Journal,  115(4),  498–522.   http://dx.doi.org10.1086/681964.   Graham,  Steve,  McKeown,  Debra,  Kiuhara,  Sharlene,  &  Harris,  Karen  R.  (2012).  A  meta-­‐analysis  of  writing   instruction  for  students  in  the  elementary  grades.  Journal  of  Educational  Psychology,  104(4),  879– 896.  http://dx.doi.org10.1037/a0029185.   Hall,  Chris  (1990).  Managing  the  complexity  of  revising  across  languages.  TESOL  Quarterly,  24(1),  43–60.   http://dx.doi.org10.2307/3586851.   Hendriks,  Henriëtte  (1998).  Reference  to  person  and  space  in  narrative  discourse:  A  comparison  of  adult   second  language  and  child  first  language  acquisition.  Studi  Italiani  di  Linguistica  Teorica  e  Applicata,   XXVII(1),  67–86.   Hickmann,  Maya  (1995).  Discourse  organisation  and  the  development  of  reference  to  person,  space,  and  time.   In  Paul  Fletcher  &  Brian  MacWhinney  (Eds.),  The  handbook  of  child  language  (pp.  194–218).  Oxford,   United  Kingdom:  Wiley-­‐Blackwell.     Hinkel,  Eli  (2001).  Matters  of  cohesion  in  L2  academic  texts.  Applied  Language  Learning,  12(2),  111–132.   Hinkel,  Eli  (2004).  Tense,  aspect  and  the  passive  voice  in  L1  and  L2  academic  texts.  Language  Teaching   Research,  8(1),  5–29.  http://dx.doi.org10.1191/1362168804lr132oa.   Hyland,  Ken  (2007).  Genre  pedagogy:  Language,  literacy  and  L2  writing  instruction.  Journal  of  Second   Language  Writing,  16,  148–164.  http://dx.doi.org10.1016/j.jslw.2007.07.005.   Jones,  Stan,  &  Tetroe,  Jacqueline  (1987).  Composing  in  a  second  language.  In  Ann  Matsuhashi  (Ed.),  Writing  in   real  time:  Modelling  production  processes  (pp.  34–57).  Norwood,  New  Jersey:  Ablex.   PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 16   Kersten,  Kristin  (2009).  Verbal  inflections  in  L2  child  narratives.  Trier,  Germany:  Wissenschaftlicher  Verlag   Trier.   Koster,  Monica,  Tribushinina,  Elena,  de  Jong,  Peter  F.,  &  van  den  Bergh,  Huub  (2015).  Teaching  children  to   write:  A  meta-­‐analysis  of  writing  intervention  research.  Journal  of  Writing  Research,  7(2),  299-­‐324.   Kuiken,  Folkert,  &  Vedder,  Ineke  (2017).  Functional  adequacy  in  L2  writing:  Towards  a  new  rating  scale.   Language  Testing,  34(3),  321–336.  http://dx.doi.org10.1177/0265532216663991   Leki,      Ilona,  Cumming,  Alister,    &  Silva,  Tony  (2006).  Second-­‐language  composition  teaching  and  learning.  In   Peter  Smagorinsky  (Ed.),  Research  on  composition:  Multiple  perspectives  on  two  decades  of  change  (pp.   141–169).  New  York,  New  York  and  London:  Teachers  College  Press.   Montrul,  Silvina  (2016).  The  acquisition  of  heritage  languages.  Cambridge,  United  Kingdom:  Cambridge   University  Press.   Oi,  Kyoko  Mizuno  (1984).  Cross-­‐cultural  differences  in  rhetorical  patterning:  A  study  of  Japanese  and  English.   Dissertation  Abstracts  International,  45(8),  2511A.   Olson,  Carol  Booth,  Scarcella,  Robin,  &  Matuchniak,  Tina  (2013).  Best  practices  in  teaching  writing  to  English   learners.  In  Steve  Graham,    Charles  A.  MacArthur,  and  Jill  Fitzgerald,  (Eds.),  Best  practices  in  writing   instruction  (pp.  381–402).  New  York,  New  York:  Guilford  Press.   Pallotti,  Gabriele  (2010).  Doing  interlanguage  analysis  in  school  contexts.  In  Inge  Bartning,  Maisa  Martin  &   Ineke  Vedder  (Eds.),  Communicative  proficiency  and  linguistic  development:  Intersections  between  SLA   and  language  testing  research  (pp.  159-­‐190).  Amsterdam:  Eurosla  Monographs.   Pallotti,  Gabriele  (2017a).  Osservare  l’interlingua:  percorsi  di  educazione  linguistica  efficace  per  ridurre  le   diseguaglianze.  In  Massimo  Vedovelli    (Ed.),  L’italiano  dei  nuovi  italiani.  Atti  del  XIX  Convegno   Nazionale  GISCEL  (pp.  505-­‐520).  Roma,  Italia:  Aracne.   Pallotti,  Gabriele  (2017b).  Applying  the  interlanguage  approach  to  language  teaching.  International  Review  of   Applied  Linguistics  in  Language  Teaching  (IRAL),  55(4),  393–412.  http://dx.doi.org10.1515/iral-­‐ 2017-­‐0145.   Pallotti,  Gabriele,  &  Rosi,  Fabiana  (2017).  Educazione  linguistica  inclusiva  nella  scuola  primaria:  percorsi  di   sperimentazione  e  analisi  dell’efficacia.  In  Loredana  Corrà  (Ed.),  Educazione  linguistica  in  classi   multietniche  (pp.  117-­‐142).  Roma,  Italia:  Aracne.   Pallotti,  Gabriele,  Borghetti,  Claudia,  &  Rosi,  Fabiana  (in  preparation).  Insegnare  a  scrivere  nella  scuola   primaria.  Percorsi  educativi  e  indagini  sperimentali.   Reid,  Joy  M.  (1988).  Quantitative  differences  in  English  prose  written  by  Arabic,  Chinese,  Spanish,  and  English   students.  Dissertation  Abstracts  International,  50(3),  672A.   Rietdijk,  Saskia,  Janssen,  Tanja,  van  Weijen,  Daphne,  van  den  Bergh,  Huub,  &  Rijlaarsdam,  Gert  (2017).   Improving  writing  in  primary  schools  through  a  comprehensive  writing  program.  The  Journal  of   Writing  Research,  9(2),  173–225.  http://dx.doi.org10.17239/jowr-­‐2017.09.02.04   Schunk,  Dale  H.,  &  Swartz,  Carl  W.  (1993).  Goals  and  progress  feedback:  Effects  on  self-­‐efficacy  and  writing   achievement.  Contemporary  Educational  Psychology,  18,  337–354.   http://dx.doi.org10.1006/ceps.1993.1024.   Selinker,  Larry  (1972).  Interlanguage.  International  Review  of  Applied  Linguistics  in  Language  Teaching,  10(2),   209–231.   Silva,  Tony  (1993).  Toward  an  understanding  of  the  distinct  nature  of  L2  writing:  The  ESL  research  and  its   implications.  TESOL  Quarterly,  27(4),  657–677.  http://dx.doi.org10.2307/3587400.   Solarino,  Rosaria  (2009).  Imparare  dagli  errori.  Napoli,  Italia:  Tecnodid.   Stevenson,  Marie,  Schoonen,  Rob,  &  de  Glopper,  Kees  (2006).  Revising  in  two  languages:  A  multi-­‐dimensional   comparison  of  online  writing  revisions  in  L1  and  FL,  Journal  of  Second  Language  Writing,  15,  201– 233,  http://dx.doi.orgdoi.org/10.1016/j.jslw.2006.06.002.   Storch,  Neomy  (2013).  Collaborative  writing  in  L2  classrooms.  Bristol,  United  Kingdom:  Multilingual  Matters.   Tracy,  Brenda,  Reid,  Robert,  &  Graham,  Steve  (2009).  Teaching  young  students  strategies  for  planning  and   drafting  stories:  The  impact  of  self-­‐regulated  strategy  development.  Journal  of  Educational  Research,   102,  323–331.  http://dx.doi.org10.3200/JOER.102.5.323-­‐332.   AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 17   Uzawa,  Kozue  (1996).  Second  language  learners’  processes  of  L1  writing,  L2  writing,  and  translation  from  L1   into  L2.  Journal  of  Second  Language  Writing,  5(3),  271–294.  http://dx.doi.org10.1016/S1060-­‐ 3743(96)90005-­‐3.   van  Waes,  Luuk,  &  Leijten,  Mariëlle  (2015).  Fluency  in  writing:  A  multidimensional  perspective  on  writing   fluency  applied  to  L1  and  L2.  Computers  and  Composition,  38(A),  79–95.   http://dx.doi.org10.1016/j.compcom.2015.09.012.   van  Weijen,  Daphne  (2009).  Writing  processes,  text  quality,  and  task  effects:  Empirical  studies  in  first  and   second  language  writing.  Utrecht,  The  Netherlands:  LOT.   White,  Ron,  &  Arndt,  Valerie  (1991).  Process  Writing.  Harlow,  United  Kingdom:  Longman. PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 18   Appendix  A RATING SCALES FOR HOLISTIC ASSESSMENT   Content: Is the number of information units provided in the text adequate and relevant? 1) The number of ideas is not at all adequate and insufficient and the ideas are unrelated to each other. 2) The number of ideas is scarcely adequate and the ideas lack consistency. 3) The number of ideas is somewhat adequate, even though they are not very consistent. 4) The number of ideas is adequate and they are sufficiently consistent. 5) The number of ideas is very adequate and they are very consistent to each other. 6) The number of ideas is absolutely adequate and they look very consistent to each other. Comprehensibility: How much effort is required to understand text purpose and ideas? 1) The text is not at all comprehensible. Ideas and purposes are unclearly stated and the efforts of the reader to understand it are ineffective. 2) The text is scarcely comprehensible, its purposes are not clearly stated and the reader struggles to understand the ideas of the writer. The reader has to guess most of the ideas and purposes. 3) The text is somewhat comprehensible, some sentences are hard to understand on a first reading. A second reading helps clarify text purposes and the ideas conveyed, but some doubts persist. 4) The text is comprehensible. Only a few sentences are unclear but are understood, without too much effort, after a second reading. 5) The text is easily comprehensible and reads smoothly. Comprehensibility is not an issue. 6) The text is very easily comprehensible and highly readable. The ideas and the purpose are clearly stated. Coherence and cohesion, 1: Is the text coherent and cohesive (e.g. cohesive devices, strategies)? 1) The text is not at all coherent. Unrelated progressions and coherence breaks are very common. The writer does not use any anaphoric device (pronouns, sentences with a clearly interpretable null subject). The text is not at all cohesive, connectives are hardly ever used, and ideas are unrelated. 2) The text is scarcely coherent. The writer often uses unrelated progressions; when coherence is achieved, it is often done through repetitions. Few anaphoric devices and some coherence breaks. The text is not very cohesive: ideas are not well linked by connectives, which are rarely used. 3) The text is somewhat coherent. Unrelated progressions and/or repetitions are frequent. More than two sentences in a row can have the same subject, even when the subject is understood. Some anaphoric devices are used. There can be few coherence breaks. The text is somewhat cohesive: some connectives are used, but they are mostly conjunctions. 4) The text is coherent. Unrelated progressions are somewhat rare, but the writer sometimes relies on unnecessary repetitions to achieve coherence. A sufficient number of anaphoric devices is used. There may be some coherence breaks. The text is cohesive. The writer makes good use of connectives, sometimes not limiting this to conjunctions. 5) The text is very coherent: when the writer introduces a new topic, it is usually done by using connectives or connective phrases. Repetitions are very infrequent. Anaphoric devices are numerous. There are no coherence breaks. The text is very cohesive and ideas are well linked by adverbial and/or verbal connectives. 6) The writer ensures extreme coherence by integrating new ideas in the text with connectives or connective phrases. Anaphoric devices are used regularly. There are few incidences of unrelated progressions and no coherence breaks. The structure of the text is extremely cohesive, thanks to a skilful use of connectives (especially linking chunks, verbal constructions, and adverbials), often used to describe relationships between ideas. Adapted from Kuiken & Vedder 2017 AN  EXPERIMENTAL  APPROACH  TO  WRITING  INSTRUCTION   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 19   Coherence and cohesion, 2 1) Can link words or groups of words with very basic linear connectors like ‘and’ or ‘then’. 2) Can link groups of words with simple connectors like ‘and’, ‘but’, and ‘because’. 3) Can use the most frequently occurring connectors to link simple sentences in order to tell a story or describe something as a simple list of points. 4) Can link a series of shorter, discrete simple elements into a connected, linear sequence of points. 5) Can use a limited number of cohesive devices to link his/her utterances into clear, coherent discourse, though there may be some ‘jumpiness’ in a long contribution. 6) Can use a variety of linking words efficiently to clearly mark the relationships between ideas. 7) Can produce clear, smoothly flowing, well-structured speech, showing controlled use of organizational patterns, connectors, and cohesive devices. 8) Can create coherent and cohesive text making full and appropriate use of a variety of organizational patterns and a wide range of cohesive devices. (Council of Europe, 2001, par. 5.2.3.1) PALLOTTI  &  BORGHETTI   E-­‐JournALL  6(1)  (2019),  pp.  1-­‐20 20   Gabriele Pallotti,  Università di Modena e Reggio Emilia gabriele.pallotti@unimore.it EN Gabriele Pallotti is a professor of language-teaching methodology at the University of Modena and Reggio Emilia. He is a member of the Executive Committee of the European Second Language Association (EuroSLA) and the editor of the EuroSLA Studies Series. His research focuses on L2 interaction and socialization, interlanguage analysis, linguistic complexity, methodology, and epistemology in applied linguistics ES Gabriele Pallotti es profesor de metodología de enseñanza de idiomas en la Universidad de Módena y Reggio Emilia. Es miembro del Comité Ejecutivo de la Asociación Europea de Segundo Idioma (EuroSLA, por sus siglas en inglés) y editor de la serie EuroSLA Studies. Su investigación se centra en la interacción y la socialización en L2, el análisis entre idiomas, la complejidad lingüística, la metodología y la epistemología en lingüística aplicada. IT Gabriele Pallotti è professore di didattica delle lingue moderne presso l’Università di Modena e Reggio Emilia. È membro del Comitato Esecutivo della European Second Language Association (EuroSLA) e direttore della serie EuroSLA Studies. La sua ricerca si concentra sull’interazione e socializzazione in L2, analisi dell’interlingua, complessità linguistica, metodologia ed epistemologia nella linguistica applicata. Claudia Borghetti,  Università di Bologna claudia.borghetti@unibo.it EN Claudia Borghetti is a research fellow in language learning and teaching at the il Dipartimento di Lingue, Letterature e Culture Moderne, Università di Bologna, Italy. Her primary research interests are intercultural language education, student mobility and multilingualism, teaching Italian as a second language, and (academic) writing. ES Claudia Borghetti es investigadora de aprendizaje y enseñanza de idiomas en el il Dipartimento di Lingue, Letterature e Culture Moderne de la Università di Bologna, Italia. Sus principales intereses de investigación son la educación intercultural de idiomas, la movilidad estudiantil y el multilingüismo, la enseñanza de italiano como segunda lengua y la escritura académica. IT Claudia Borghetti è ricercatrice di apprendimento ed insegnamento delle lingue moderne presso il Dipartimento di Lingue, Letterature e Culture Moderne dell’Università di Bologna, in Italia. I suoi principali interessi di ricerca sono l’insegnamento interculturale delle lingue, la mobilità degli studenti e il multilinguismo, l'insegnamento dell'italiano come lingua seconda e la scrittura (accademica).